Initialise all metics

ragas/metrics/__init__.py

@@ -1,17 +1,25 @@
 from ragas.metrics.base import Evaluation, Metric
-from ragas.metrics.factual import EntailmentScore
-from ragas.metrics.similarity import SBERTScore
-from ragas.metrics.simple import BLUE, EditDistance, EditRatio, Rouge1, Rouge2, RougeL
+from ragas.metrics.factual import entailment_score, q_square
+from ragas.metrics.similarity import bert_score
+from ragas.metrics.simple import (
+    bleu_score,
+    edit_distance,
+    edit_ratio,
+    rouge1,
+    rouge2,
+    rougeL,
+)
 
 __all__ = [
     "Evaluation",
     "Metric",
-    "EntailmentScore",
-    "SBERTScore",
-    "BLUE",
-    "EditDistance",
-    "EditRatio",
-    "RougeL",
-    "Rouge1",
-    "Rouge2",
+    "entailment_score",
+    "bert_score",
+    "q_square",
+    "bleu_score",
+    "edit_distance",
+    "edit_ratio",
+    "rouge1",
+    "rouge2",
+    "rougeL",
 ]

ragas/metrics/factual.py

@@ -218,7 +218,7 @@ def __post_init__(
 
     @property
     def name(self):
-        return "Q^2"
+        return "Qsquare"
 
     @property
     def is_batchable(self):
@@ -340,5 +340,5 @@ def score(self, ground_truth: list[str], generated_text: list[str], **kwargs):
         return scores
 
 
-ENTScore = EntailmentScore()
-Q2Score = Qsquare()
+entailment_score = EntailmentScore()
+q_square = Qsquare()

ragas/metrics/similarity.py

@@ -12,12 +12,12 @@
 if t.TYPE_CHECKING:
     from torch import Tensor
 
-SBERT_METRIC = t.Literal["cosine", "euclidean"]
+BERT_METRIC = t.Literal["cosine", "euclidean"]
 
 
 @dataclass
-class SBERTScore(Metric):
-    similarity_metric: t.Literal[SBERT_METRIC] = "cosine"
+class BERTScore(Metric):
+    similarity_metric: t.Literal[BERT_METRIC] = "cosine"
     model_path: str = "all-MiniLM-L6-v2"
     batch_size: int = 1000
 
@@ -28,7 +28,7 @@ def __post_init__(self):
     def name(
         self,
     ):
-        return f"SBERT_{self.similarity_metric}"
+        return f"BERTScore_{self.similarity_metric}"
 
     @property
     def is_batchable(self):
@@ -64,4 +64,4 @@ def score(
         return score
 
 
-__all__ = ["SBERTScore"]
+bert_score = BERTScore()

ragas/metrics/simple.py

@@ -91,9 +91,9 @@ def score(self, ground_truth: t.List[str], generated_text: t.List[str]):
         return score
 
 
-Rouge1 = ROUGE("rouge1")
-Rouge2 = ROUGE("rouge2")
-RougeL = ROUGE("rougeL")
-BLUE = BLEUScore()
-EditDistance = EditScore("distance")
-EditRatio = EditScore("ratio")
+rouge1 = ROUGE("rouge1")
+rouge2 = ROUGE("rouge2")
+rougeL = ROUGE("rougeL")
+bleu_score = BLEUScore()
+edit_distance = EditScore("distance")
+edit_ratio = EditScore("ratio")

tests/benchmarks/benchmark.py

@@ -1,36 +1,27 @@
 import typing as t
 
-from datasets import Dataset, load_dataset
+from datasets import Dataset, arrow_dataset, load_dataset
 from torch.cuda import is_available
 from tqdm import tqdm
 from utils import print_table, timeit
 
-from ragas.metrics import (
-    EditDistance,
-    EditRatio,
-    EntailmentScore,
-    Evaluation,
-    Rouge1,
-    Rouge2,
-    RougeL,
-    SBERTScore,
-)
+from ragas.metrics import Evaluation, edit_distance, edit_ratio, rouge1, rouge2, rougeL
 
 DEVICE = "cuda" if is_available() else "cpu"
 BATCHES = [0, 1]
-# init metrics
-sbert_score = SBERTScore(similarity_metric="cosine")
-entail = EntailmentScore(max_length=512, device=DEVICE)
+
 METRICS = {
-    "Rouge1": Rouge1,
-    "Rouge2": Rouge2,
-    "RougeL": RougeL,
-    "EditRatio": EditRatio,
-    "EditDistance": EditDistance,
+    "Rouge1": rouge1,
+    "Rouge2": rouge2,
+    "RougeL": rougeL,
+    "EditRatio": edit_ratio,
+    "EditDistance": edit_distance,
     # "SBERTScore": sbert_score,
     # "EntailmentScore": entail,
 }
 DS = load_dataset("explodinggradients/eli5-test", split="test_eli5")
+assert isinstance(DS, arrow_dataset.Dataset), "Not an arrow_dataset"
+DS = DS.select(range(100))
 
 
 def setup() -> t.Iterator[tuple[str, Evaluation, Dataset]]: